RT @abk_tau: DeciMamba, the first context extension method for Mamba, is accepted to #ICLR2025! 🎉. New revision with more long-context resu….

RT @NimrodShabtay: I am happy to share that LiveXiv accepted to ICLR 2025 🥳.

Join millions who have switched to Grok.

RT @NimrodShabtay: Introducing LiveXiv, a new, challenging and maintainable scientific multi-modal live dataset.Paper: .

RT @RGiryes: Excited to announce that our paper on efficient diffusion models has been accepted to NeurIPS! Huge kudos to @ShadyAbh for the….

RT @abk_tau: @KyeGomezB @ItamarZimerman @ShadyAbh @nadavcohen @amirgloberson @liorwolf @RGiryes Code is now available:..

(6/6) .* Check out our code and explore UDPM for your projects: 💻🔗. * Read the full paper here: 📄🔬. #AI #MachineLearning #ImageGeneration #Research #DeepLearning #DiffusionModels #FastGeneration.

(5/6) .* UDPM has an interpretable and interpolable latent space, making it superior to traditional DDPMs in various applications. 🌐🔍.* UDPM has smaller latent variables, hence one may modify the noise maps of each step and study its effect on the generated image:

(4/6) .UDPM achieves an FID score of 6.86, 7.10, and 7.41 on the popular CIFAR10, AFHQv2, and FFHQ datasets respectively, outperforming current state-of-the-art non-distillation diffusion models; while using much fewer computational resources. 🔥📈

(3/6) .* Traditional DDPMs are powerful but computationally intensive. * UDPM progressively subsamples the latent variable, which significantly reduces the required computations. * UDPM uses 3 upsampling diffusion steps; which is equivalent to 0.3 denoising diffusion steps! 🖼️💡

(2/6) .How it works: . * In the forward process, UDPM gradually downsamples the latent variable and adds noise. * In the reverse process a network is trained to denoise and upsample the latent variables, producing high-fidelity images.

(1/6) 🌟 New Paper! 🌟. Is it possible to sample images from a diffusion model using less than one step?. Answer: Yes! ➡️ We present UDPM, a novel approach to image generation that outperforms existing models in efficiency and quality. 🚀📊.

RT @abk_tau: New Work! 🐍.What prevents Mamba from extrapolating to sequences that are significantly longer than those it was trained on? .F….

RT @CIGX: UDPM: Upsampling Diffusion Probabilistic Models.

RT @semisance: UDPM: Upsampling Diffusion Probabilistic Models.

RT @jfischoff: I like the idea behind "Upsampling Diffusion Probabilistic Models" (. Instead using a fixed sized….